A 15-year project to develop less invasive methods to diagnose pneumocystis pneumonia and to predict responses to therapy, includes the collection of oral washes, induced sputum, and bronchoalveolar lavage from patients with immunosuppressive diseases and respiratory syndromes. Samples for control patients are being collected as well. First, a polymerase chain reaction technique using a unique major surface glycoprotein (MSG) primer is being used in conjunction with a published primer to develop a highly specific and sensitive method adaptable to clinical laboratories. It is hoped that oral wash can replace sputum as the sample of choice. Second, mutations associated with drug resistance are being assessed in all organisms identified to determine the epidemiology and clinical importance of such mutations. Third, markers of strain variation are being assessed to elucidate pathogen epidemiology.